The answer is Angles

Let me simplify.

When you are taking that kind of photo, the water is a horizontal plane, the camera is pointing horizontally and near the water.

Let me have an imaginary rope between the camera and the light. Let's call this the "line of sight rope" or "rope" for short.

Imagine now that we draw a grid, like a perspective grid on the water. Separating it into squares. What we would see is that even if the squares remain horizontally, the reflection already covers more squares under the "rope", than to the sides of it.

We would need to strongly rotate these lateral squares, to the inside in order to have a reflection of the light. Any rotation to the outside would produce no reflection at all.

But only slight variations in angles, on the tiles under the rope.

So that is the answer. We only need tiny waves to have the reflection under the rope to be noticeable. When we increase the distortion on the surface we start to see the with of the reflection increase, but this already has increased the vertical length of the reflection.

The lower the light is on the horizon, the more and more tiles under the rope will be part of the reflection. Kilometers and kilometers, but almost no change on the sides.

The answer is Angles

Let me simplify with a 3D simulation of a softbox over a glossy blue plane.

In our original scenario, the water is in complete calm and we have a reflection of the same size and shape as the light.

Let me have an imaginary rope between the camera and the light. Let's call this the "line of sight rope" or "rope" for short.

Imagine now that we draw a grid, like a perspective grid on the water. Separating it into squares. What we would see is that even if the squares remain horizontally, the reflection already covers more squares under the "rope", than to the sides of it.

We would need to strongly rotate these lateral squares, to the inside in order to have a reflection of the light. Any rotation to the outside would produce no reflection at all.

But we need only slight variations in angles, on the tiles under the rope.

So that is the answer. We only need tiny waves to have the reflection under the rope to be noticeable. When we increase the distortion on the surface we start to see the with of the reflection increase, but this already has increased the vertical length of the reflection.

The lower the light is on the horizon, the more and more tiles under the rope will be part of the reflection. Kilometers and kilometers, but almost no change on the sides.

The answer is Angles

Let me simplify.

When you are taking that kind of photo, the water is a horizontal plane, the camera is pointing horizontally and near the water.

Let me have an imaginary rope between the camera and the light. Let's call this the "line of sight rope" or "rope" for short.

Imagine now that we draw a grid, like a perspective grid on the water. Separating it into squares. What we would see is that even if the squares remain horizontally, the reflection already covers more squares under the "rope", than to the sides of it.

We would need to strongly rotate these lateral squares, to the inside in order to have a reflection of the light. Any rotation to the outside would produce no reflection at all.

But only slight variations in angles, on the tiles under the rope.

So that is the answer. We only need tiny waves to have the reflection under the rope to be noticeable. When we increase the distortion on the surface we start to see the with of the reflection increase, but this already has increased the vertical length of the reflection.

The answer is Angles

Let me simplify.

When you are taking that kind of photo, the water is a horizontal plane, the camera is pointing horizontally and near the water.

Let me have an imaginary rope between the camera and the light. Let's call this the "line of sight rope" or "rope" for short.

Imagine now that we draw a grid, like a perspective grid on the water. Separating it into squares. What we would see is that even if the squares remain horizontally, the reflection already covers more squares under the "rope", than to the sides of it.

We would need to strongly rotate these lateral squares, to the inside in order to have a reflection of the light. Any rotation to the outside would produce no reflection at all.

But only slight variations in angles, on the tiles under the rope.

So that is the answer. We only need tiny waves to have the reflection under the rope to be noticeable. When we increase the distortion on the surface we start to see the with of the reflection increase, but this already has increased the vertical length of the reflection.

The lower the light is on the horizon, the more and more tiles under the rope will be part of the reflection. Kilometers and kilometers, but almost no change on the sides.